Altman responded to users' complaints again and announced that the inference quota for Plus users of GPT - 5 would be increased to 3,000 times per week.

On August 11th, it was reported that Sam Altman, the CEO of OpenAI, shared a series of the latest updates and personal thoughts about the launch of GPT-5 on the X platform late at local time. Altman also said that the usage limit of the reasoning model for ChatGPT Plus users will be increased to 3,000 times per week, and they will also get a very small number of Pro reasoning times. This adjustment is directly due to the surge in user usage.

Meanwhile, xAI, an artificial intelligence company under Elon Musk, announced that the basic version of its latest AI model, Grok 4, is now available for free to users worldwide. The usage limit for ordinary users is 5 requests every 12 hours. The subscription fee for Grok 4 is $30 per month, and the more powerful Grok 4 Heavy version costs $300 per month. In this new round of product releases, both companies are competing to offer more benefits in an effort to gain the upper hand in the AI field.

Users' Crazy Complaints After the Launch of GPT-5

After GPT-5 was launched at around midnight on August 8th Beijing time, it quickly sparked extensive discussions. Altman admitted that although GPT-5 performs better in most aspects, the company underestimated users' fondness for certain features of GPT-4o. For example, users have very different views on the relative advantages of GPT-4o and GPT-5.

The usage limit is a major issue that users have been complaining about. In response to user feedback, OpenAI has brought good news to ChatGPT Plus users: the usage limit of the reasoning model will be significantly increased.

Altman revealed that the proportion of daily active users of the reasoning model has increased significantly: for free users, it has soared from less than 1% to 7%, and for Plus users, it has skyrocketed from 7% to 24%. This data not only confirms the popularity of the model but also addresses netizens' "complaints".

Users' Attachment to AI Models and Considerations for Mental Health

On the morning of August 11th, Altman shared his in - depth thoughts on users' interaction with AI on the X platform. He observed that users' attachment to specific AI models is far greater than to previous technologies, and this emotional connection is stronger and appears suddenly. Therefore, suddenly abandoning the old models that users rely on is a mistake. The company has been closely monitoring this phenomenon for the past year, but it has not yet attracted mainstream attention.

In the field of mental health, Altman attaches great importance to the value of responsibility. He noticed that people sometimes use technology, including artificial intelligence, in self - harming ways. When users are in a mentally vulnerable state and prone to delusions, AI should not be a "booster" for such negative tendencies.

Altman said that many people now regard ChatGPT as a "therapist" or a "life coach", and such interactions often bring positive meaning. If users can get useful advice from it, achieve their goals, and improve their life satisfaction, even if they become highly dependent on it, it is still reassuring.

On the contrary, if such interactions only bring short - term pleasure but damage long - term happiness, it becomes a problem that needs to be addressed. Altman imagines that in the future, billions of users will trust AI and let it participate in major decisions. Although the potential behind this is huge, it also makes people feel a little uneasy. Therefore, both society and OpenAI must work together to ensure that it has completely positive effects.

This article is from "Tencent Technology", author: Wuji. It is published by 36Kr with authorization.